TheBackShed.com

Forum Index : Microcontroller and PC projects : Picomite(VGA) V5.07.07 betas - bug fixes + focus on PIO

Page 4 of 16

Author

Message

disco4now

Guru

Joined: 18/12/2014
Location: Australia
Posts: 900

Posted: 03:19am 20 Jan 2023

Copy link to clipboard

Print this post

On Beta 6 the example from page 2 doesn't seem the work properly if it is ask to transfer more that 32 words/items?.

This outputs to the LEDs as expected.
PIO DMA_OUT 1,2,32,packed%(),dmadone,16,32

Anything where nbr is >32 seems to not wrap to the start of the packed%()
as expected and LEDs are not as expected after the first two cycles e.g.
PIO DMA_OUT 1,2,64,packed%(),dmadone,16,32
PIO DMA_OUT 1,2,&H100,packed%(),dmadone,16,32
PIO DMA_OUT 1,2,&H1000,packed%(),dmadone,16,32

Also a documentation question?
'PIO DMA_OUT pio, sm, nbr, data%() [,completioninterrupt] [,transfersize] [,loopbackcount]
nbr specifies how many 32-bit words to transfer (as in the Readme.txt)
nbr specifies how many data items are to be read or written (should it be this?)

matherp said Picomite(VGA) V5.07.07b5

' demo program to demonstrate the use of the PIO interrupt
' the PIO is programmed to pull 1 word from FIFO at every rising edge of the GP pin.
' the LSB 5 bits are shown on GP1..GP5 (diagnostics)
' the GP0 pin is controlled from MMBasic in this demo, but could also be driven from
' an external source.

'-------------------------------- PIO state machine -----------------------------------

'PIO clear, and attach pins to PIO 1
PIO clear 1
SetPin gp1,pio1
SetPin gp2,pio1
SetPin gp3,pio1
SetPin gp4,pio1
SetPin gp5,pio1
Dim integer unpacked%(31),packed%(15)
Dim integer packedadd=Peek(varaddr packed%())
Dim integer unpackedadd=Peek(varaddr unpacked%())
For i=0 To 31
unpacked%(i)=i
Next
Memory pack unpackedadd,packedadd,32,16

'the PIO program
'address � �code � �mnemonics � � � � � � comment
' �0 � � �E09F � � �SET pindirs &b11111 � GP1..GP5 out
' �1 � � �2000 � � �WAIT GP0=0 � � � � � �wait for external trigger on GP0 to pull
' �2 � � �2080 � � �WAIT GP0=1
' �3 � � �8080 � � �PULL noblock � � � � �pulls data off fifo
' �4 � � �6005 � � �OUT OSR > pins 5 � � �shift 5 bits to GP1..5 pins
' �5 � � �0001 � � �JMP 1 � � � � � � � � repeat

PIO program line 1,0,&hE09F
PIO program line 1,1,&h2000
PIO program line 1,2,&h2080
PIO program line 1,3,&h8080
PIO program line 1,4,&h6005
PIO program line 1,5,&h0001

'write the configuration, running 1MHz
f=1e6
p=Pio(pinctrl 0,5,5,gp0,,gp1,gp1) � �'pins GP1...GP5 = out, GP0 = in
' �s=PIO(shiftctrl 0,0,0,0,1,1) � � � �'default value
' �e=PIO(execctrl gp0,0,31) � � � � � �'default value: wrap defaults, GP0 = cond JMP pin
' �PIO init machine 1,0,f,p,e,s,0 � � �'PIO 1 SM 0 start adress = 0
PIO init machine 1,2,f,p,,,0 � � � � 'PIO 1 SM 0 start adress = 0

'-------------------------------- MMBasic MAIN ---------------------------------------

'Halt the PIO from GP0 from MMBasic before it fills the FIFO out of control
SetPin gp0,dout
Pin(gp0)=0

'variables
Dim d%=0

'start the pio1 code, state machine 1.0 will be waiting for a rising edge on the GP0 pin
'then it will pull 1 word from fifo. Then it will wait for the falling edge. The cycle repeats.

'set the dma to loop over 32 16-bit shorts for a very long time

PIO DMA_OUT 1,2,&H10000,packed%(),dmadone,16,32

'freguently PIO pull data from the fifo
SetTick 200,PullFifo

'Main: interrupt free loop polling the NOT TX_FULL flag in FSTAT
Do
If a% Then
� SetTick 0,0
� End
EndIf
Loop

End

'----------------------------------- SUBS ---------------------------------------

'in this sub we toggle the GP0 pin to force the PIO to fetch data from the FIFO
Sub PullFifo
Pulse gp0,10
End Sub

Sub dmadone
Print "All done"
a%=1
End Sub

Latest F4 Latest H7

phil99

Guru

Joined: 11/02/2018
Location: Australia
Posts: 2141

Posted: 04:49am 20 Jan 2023

Copy link to clipboard

Print this post

Also the last version of the 3 phase program for Beta 6 on page 3 crashes VGA beta 6.1 and self erases.

Quote Here is your code with the mod to make it reliable (call init in the interrupt routine)

' Test signal: Three phase motor drive - picomite V50706+
' runs autonomously on GP0...GP5 50Hz

Saved 2593 bytes
> RUN
press any key to trigger
yes
Error: Invalid address - resetting
PicoMiteVGA MMBasic Version 5.07.07b6
Copyright 2011-2023 Geoff Graham
Copyright 2016-2023 Peter Mather

> LIST
>
> option list
PicoMiteVGA MMBasic Version 5.07.07b6
OPTION COLOURCODE ON
OPTION KEYBOARD US
OPTION DISPLAY 60, 106
OPTION SDCARD GP13, GP10, GP11, GP12
OPTION COUNT GP6,GP7,GP14,GP15
OPTION DEFAULT FONT 7, 1
>

Re-loaded program from TBS same result, re-flashed b6.1 same result.
Copied this version from non-VGA Pico and it works. Trivial changes to PWM section that should have no effect.

Edited 2023-01-20 15:27 by phil99

Pluto
Guru

Joined: 09/06/2017
Location: Finland
Posts: 359

Posted: 05:25am 20 Jan 2023

Copy link to clipboard

Print this post

Matherp said Should be OK now, please re-download. No version change

This corrected the MEMORY UNPACK issue (as you probably have tested and noticed yourself already).

Volhout
Guru

Joined: 05/03/2018
Location: Netherlands
Posts: 4256

Posted: 08:14am 20 Jan 2023

Copy link to clipboard

Print this post

Hi Peter,

I have a few questions regarding the DMA. Not that it does not work, but I am not sure how to set it up correctly.

1/ Create ring buffer
Anything smaller than 256 does not seem to work (32, 64, 256) despite being a power of 2

2/ PIO DMA_IN pio, sm, nbr, data%() [,completioninterrupt] [,transfersize] [,loopbackcount]
When you use a ring buffer, do you need to feed it with "nbr" ?

3/ PIO DMA_IN pio, sm, nbr, data%() [,completioninterrupt] [,transfersize] [,loopbackcount]
When you use a ring buffer, does it ever reach "completioninterrupt" ?

I hope I get this right

'The DMA should take (2*length 32 bit) samples, this takes up (1*length 64 bit) ring buffer
'It should never trigger the ReadyInt interrupt (ringbuffer). Not sure if we need specify
'the number of samples.
PIO DMA_IN 1,0,2*length%,packed%(),ReadyInt,32,length% 'start DMA with ring buffer

I tried loopback parameter:

length
2*length
4*length
8*length (number of bytes the DMA buffer is in size)
ERROR: Data alignment error

length-1
2*length-1
4*length-1
8*length-1 (number of bytes the DMA buffer is in size)
ERROR: not a power of 2

I have no clue what to do next...

Volhout

' Test signal: Three phase motor drive - picomite V50706+
' runs autonomously on GP0...GP5 50Hz

SetPin gp0,pwm 'CH 0a
SetPin gp1,pwm 'CH 0b
SetPin gp2,pwm 'CH 1a
SetPin gp3,pwm 'CH 1b
SetPin gp4,pwm 'CH 2a
SetPin gp5,pwm 'CH 2b

PWM 0, 50, 33.33, -66.66, 1, 1
PWM 1, 50, 33.33, -66.66, 1, 1
PWM 2, 50, 33.33, -66.66, 1, 1

PWM sync 0, 100/3, 200/3
Pause 20

'----------------------------------- LA code PIO --------------------------

'PIO code to sample GP0..GP6 for logic analyzer
PIO clear 1

'in this program the PIO reads GP0..GP5 brute force
'and pushes data into FIFO. The clock speed determines the
'sampling rate. There are 3 instructions per cycle
'taking 10000/3 / 50 = 66 samples per cycle.

'the PIO program
'address code mnemonics comment
'.wrap target
' 0 80A0 pull block 'pull data from FIFO into OSR when available
' 1 A027 mov(x,osr) 'load gate time (in clock pulses) from osr

' 2 4006 IN pins 6 'read 6 pins
' 3 8000 PUSH noblock 'and push to FIFO
' 4 00C2 JMP (GP22) 2 'while no trigger (GP22=1) loop to 2

' 5 4006 IN pins 6 'read 6 pins
' 6 8000 PUSH noblock 'and push to FIFO
' 7 0045 JMP (X<>0) 5 X-- 'while X<>0 decrement X and jump to 5
'.wrap

'program pio1
PIO program line 1,0,&h80A0
PIO program line 1,1,&hA027
PIO program line 1,2,&h4006
PIO program line 1,3,&h8000
PIO program line 1,4,&h00C2
PIO program line 1,5,&h4006
PIO program line 1,6,&h8000
PIO program line 1,7,&h0045

'configuration
f=1e4 'PIO run at 10kHz
p=Pio(pinctrl 0,0,0,gp0,,,) 'IN base = GP0
e=Pio(execctrl gp22,0,7) 'wrap 7 to 0, gp22 is conditional JMP pin
s=Pio(shiftctrl 0,0,0,0,0,0) 'shift in through LSB, out is not used

'write the configuration, running 10kHz
PIO init machine 1,0,f,p,e,s,0 'start address = 0

'---------------------------- LA code MMBasic ----------------------------------
'prepare trigger signal set to idle (above PIO code needs falling edge trigger)
SetPin gp22,dout
Pin(gp22)=1

'do not start PIO, the DMA will do that.

'reserve data buffers
length%=256
Dim data%(2*length%-1) 'array to put the 32 bit samples
'Dim packed%(length%-1) 'DMA array to pack 32 bit samples in 64 bit integers
Dim packed% 'name of the array to be formed as circular array
PIO MAKE RING BUFFER packed%,length% 'this creates array called "packed%()"

'dim variables
Dim a$(1)=("_","-")
Dim done%=0

'below code must be adapted for circular FIFO buffer........................!!!!

Do
Print "press <SPACE> to arm LA"
Do : Loop Until Inkey$=" "

'The DMA should take (2*length 32 bit) samples, this takes up (1*length 64 bit) ring buffer
'It should never trigger the ReadyInt interrupt (ringbuffer). Not sure if we need specify
'the number of samples.
'PIO DMA_IN 1,0,2*length%,packed%(),ReadyInt,32,length% 'start DMA with ring buffer
PIO DMA_IN 1,0,2*length%,packed%(),ReadyInt,,4*length%-1 'start DMA with ring buffer

'write to PIO the number of post trigger samples, this starts the logging of pre-trigger data
PIO WRITE 1,0,1,length%

Pause 200

Print "press "t" to trigger LA"
Do : Loop Until Inkey$="t"

'the trigger pulse, must be wider that 3 / f(pio) in seconds.
'for 10kHz, that is 0.3ms
Pulse gp22,1 'pulse 1ms

'from here the PIO should autocomplete and wait for another post trigger length% written

'wait a bit (can calculate from post trigger length and sampling speed), default for now
Pause 200

'stop DMA, stop PIO, re-config PIO
PIO DMA_IN OFF
PIO STOP 1,0
PIO init machine 1,0,f,p,e,s,0 'start address = 0

'process the data from the ring buffer
Memory unpack packed%(),data%(),2*length%,32
For j=0 To 5
mask%=2^j
For i=0 To 2*length%-1
Print a$(((data%(i) And mask%)=mask%));
Next i
Print
Next j
Math set 0,packed%()

Loop

End

'-----------------------------------SUBS MMBasic --------------------------------
Sub ReadyInt
Print "we should not get here ------------- *****"
PIO DMA_IN OFF
End Sub

PicomiteVGA PETSCII ROBOTS

matherp
Guru

Joined: 11/12/2012
Location: United Kingdom
Posts: 9139

Posted: 08:35am 20 Jan 2023

Copy link to clipboard

Print this post

Quote 1/ Create ring buffer
�Anything smaller than 256 does not seem to work (32, 64, 256) despite being a power of 2

Quote All MMBasic arrays are aligned to a 256 byte boundary but to create an array which is guaranteed to be aligned to a 512 byte boundary or greater a new sub command is used

Therefore you can just use a normal array for anything less than or equal to 256 bytes. The size of the ring buffer is in bytes so you adjust this to your transfer size (1,2,or 4 bytes) and the number of transfers in the loop (loopbackcount)

Quote 2/ PIO DMA_IN pio, sm, nbr, data%() [,completioninterrupt] [,transfersize] [,loopbackcount]
When you use a ring buffer, do you need to feed it with "nbr" ?

Yes, the RP2040 doesn't have a loop-forever capability so nbr needs to be specified. You can use &HFFFFFFFF

Quote 3/ PIO DMA_IN pio, sm, nbr, data%() [,completioninterrupt] [,transfersize] [,loopbackcount]
When you use a ring buffer, does it ever reach "completioninterrupt" ?

Yes: when nbr runs out

'dma_out continuous
' demo program to demonstrate the use of the PIO interrupt
' the PIO is programmed to pull 1 word from FIFO at every rising edge of the GP pin.
' the LSB 5 bits are shown on GP1..GP5 (diagnostics)
' the GP0 pin is controlled from MMBasic in this demo, but could also be driven from
' an external source.

'-------------------------------- PIO state machine -----------------------------------

'PIO clear, and attach pins to PIO 1
PIO clear 1
SetPin gp1,pio1
SetPin gp2,pio1
SetPin gp3,pio1
SetPin gp4,pio1
SetPin gp5,pio1
Dim integer unpacked%(127),packed%
npackedadd=Peek(varaddr unpacked%())
For i=0 To 31
unpacked%(i)=i
unpacked%(63-i)=i
unpacked%(64+i)=i
unpacked%(127-i)=i
Next
PIO make ring buffer packed%,512
Memory pack unpacked%(),packed%(),128,32

'the PIO program
'address � �code � �mnemonics � � � � � � comment
' �0 � � �E09F � � �SET pindirs &b11111 � GP1..GP5 out
' �1 � � �2000 � � �WAIT GP0=0 � � � � � �wait for external trigger on GP0 to pull
' �2 � � �2080 � � �WAIT GP0=1
' �3 � � �8080 � � �PULL noblock � � � � �pulls data off fifo
' �4 � � �6005 � � �OUT OSR > pins 5 � � �shift 5 bits to GP1..5 pins
' �5 � � �0001 � � �JMP 1 � � � � � � � � repeat

PIO program line 1,0,&hE09F
PIO program line 1,1,&h2000
PIO program line 1,2,&h2080
PIO program line 1,3,&h8080
PIO program line 1,4,&h6005
PIO program line 1,5,&h0001

'write the configuration, running 1MHz
f=1e6
p=Pio(pinctrl 0,5,5,gp0,,gp1,gp1) � �'pins GP1...GP5 = out, GP0 = in
' �s=PIO(shiftctrl 0,0,0,0,1,1) � � � �'default value
' �e=PIO(execctrl gp0,0,31) � � � � � �'default value: wrap defaults, GP0 = cond JMP pin
' �PIO init machine 1,0,f,p,e,s,0 � � �'PIO 1 SM 0 start adress = 0
PIO init machine 1,2,f,p,,,0 � � � � 'PIO 1 SM 0 start adress = 0

'-------------------------------- MMBasic MAIN ---------------------------------------

'Halt the PIO from GP0 from MMBasic before it fills the FIFO out of control
SetPin gp0,dout
Pin(gp0)=0

'variables
Dim d%=0

'start the pio1 code, state machine 1.0 will be waiting for a rising edge on the GP0 pin
'then it will pull 1 word from fifo. Then it will wait for the falling edge. The cycle repeats.

'set the dma to loop over 32 16-bit shorts for a very long time
PIO DMA_OUT 1,2,&H10000,packed%(),dmadone,32,128

'freguently PIO pull data from the fifo
SetTick 200,PullFifo

'Main: interrupt free loop polling the NOT TX_FULL flag in FSTAT
Do
If a% Then
�SetTick 0,0
�End
EndIf
Loop

End

'----------------------------------- SUBS ---------------------------------------

'in this sub we toggle the GP0 pin to force the PIO to fetch data from the FIFO
Sub PullFifo
Pulse gp0,10
End Sub

Sub dmadone
Print "All done"
a%=1
End Sub

Edited 2023-01-20 18:46 by matherp

Volhout
Guru

Joined: 05/03/2018
Location: Netherlands
Posts: 4256

Posted: 09:25am 20 Jan 2023

Copy link to clipboard

Print this post

@Peter

Thanks for educating. Tha basis works now. But I notice that when you force stop the DMA, you trigger DMA_Ready interrupt. I can work around that.

Also the "nbr" is a signed integer (&hFFFFFFFF does not work, it needs to be &h7FFFFFFF).

Volhout
Edited 2023-01-20 19:26 by Volhout

PicomiteVGA PETSCII ROBOTS

matherp
Guru

Joined: 11/12/2012
Location: United Kingdom
Posts: 9139

Posted: 09:35am 20 Jan 2023

Copy link to clipboard

Print this post

Quote But I notice that when you force stop the DMA, you trigger DMA_Ready interrupt.

If you don't want the interrupt don't include it in the command

PIO DMA_OUT 1,2,&H10000,packed%(),,32,128

Volhout
Guru

Joined: 05/03/2018
Location: Netherlands
Posts: 4256

Posted: 09:36am 20 Jan 2023

Copy link to clipboard

Print this post

Getting there, need to re-align trigger to the centre. All excited !!!
Trigger is parallel wired to channel 15.

Edited 2023-01-20 19:37 by Volhout

PicomiteVGA PETSCII ROBOTS

Pluto
Guru

Joined: 09/06/2017
Location: Finland
Posts: 359

Posted: 10:53am 20 Jan 2023

Copy link to clipboard

Print this post

From Peter's program above:

For i=0 To 31
unpacked%(i)=i
unpacked%(63-i)=i
unpacked%(64+i)=i
unpacked%(127-i)=i
Next
PIO make ring buffer packed%,512
Memory pack unpacked%(),packed%(),128,32

When I run the program and look at the outputs on the LEDS (GP1 to GP5):

Sub PullFifo
Pulse gp0,10
inc count
Leds=pin(gp1)+(pin(GP2)<<1)+(pin(gp3)<<2)+(pin(gp4)<<3)+(pin(gp5)<<4)
print count;" &h";HEX$(count)," GP1:5:";Leds
End Sub

Running the program all seems to be OK in the beginning, but after 128 PullFifo the output on the leds become strange (in my mind):

113 &h71 GP1:5: 15
114 &h72 GP1:5: 14
115 &h73 GP1:5: 13
116 &h74 GP1:5: 12
117 &h75 GP1:5: 11
118 &h76 GP1:5: 10
119 &h77 GP1:5: 9
120 &h78 GP1:5: 8
121 &h79 GP1:5: 7
122 &h7A GP1:5: 6
123 &h7B GP1:5: 5
124 &h7C GP1:5: 4
125 &h7D GP1:5: 3
126 &h7E GP1:5: 2
127 &h7F GP1:5: 1
128 &h80 GP1:5: 0
129 &h81 GP1:5: 0 Extra zeroes starting from here.
130 &h82 GP1:5: 0
131 &h83 GP1:5: 1
132 &h84 GP1:5: 0
133 &h85 GP1:5: 2
134 &h86 GP1:5: 0
135 &h87 GP1:5: 3
136 &h88 GP1:5: 0
137 &h89 GP1:5: 4
138 &h8A GP1:5: 0
139 &h8B GP1:5: 5
140 &h8C GP1:5: 0
141 &h8D GP1:5: 6
142 &h8E GP1:5: 0

After a while nothing is output anymore. It is quite possible that I have misunderstood the Ringbuffer, but to me it looks suspicious.

Volhout
Guru

Joined: 05/03/2018
Location: Netherlands
Posts: 4256

Posted: 11:21am 20 Jan 2023

Copy link to clipboard

Print this post

Just to give you something to play with, no questions anymore. The core works. From here is is elbow grease to make it into a logic analyzer.

ToDo:
- Decide on trigger logic
- Make pan and scan work
- Nice VGA graphics (the first will be a 6 channel fast LA) maybe later a 32 channel slower one, with different trigger logic. Requires external HW.

Here is the working core.

'logic analyzer framework research program
'samples GP0...GP15 with triggering on GP22 on standard picomite
'uses PIO and ring buffer under DMA.
'requires Picomite MMBasic V50707b6 or newer.
'Volhout 20-1-2023

' Test signal: Three phase motor drive - picomite V50706+
' runs autonomously on GP0...GP5 50Hz

SetPin gp0,pwm 'CH 0a
SetPin gp1,pwm 'CH 0b
SetPin gp2,pwm 'CH 1a
SetPin gp3,pwm 'CH 1b
SetPin gp4,pwm 'CH 2a
SetPin gp5,pwm 'CH 2b

PWM 0, 50, 33.33, -66.66, 1, 1
PWM 1, 50, 33.33, -66.66, 1, 1
PWM 2, 50, 33.33, -66.66, 1, 1

PWM sync 0, 100/3, 200/3
Pause 20

'----------------------------------- LA code PIO --------------------------

'PIO code to sample GP0..GP15 for logic analyzer
PIO clear 1

'in this program the PIO reads GP0..GP5 brute force
'and pushes data into FIFO. The clock speed determines the
'sampling rate. There are 3 instructions per cycle
'taking 10000/3 / 50 = 66 samples per cycle.

'the PIO program
'address code mnemonics comment
'.wrap target
' 0 80A0 pull block 'pull data from FIFO into OSR when available
' 1 A027 mov(x,osr) 'load gate time (in clock pulses) from osr

' 2 4010 IN pins 16 'read 16 pins
' 3 8000 PUSH noblock 'and push to FIFO
' 4 00C2 JMP (GP22) 2 'while no trigger (GP22=1) loop to 2

' 5 4010 IN pins 16 'read 16 pins
' 6 8000 PUSH noblock 'and push to FIFO
' 7 0045 JMP (X<>0) 5 X-- 'while X<>0 decrement X and jump to 5
'.wrap

'program pio1
PIO program line 1,0,&h80A0
PIO program line 1,1,&hA027
PIO program line 1,2,&h4010
PIO program line 1,3,&h8000
PIO program line 1,4,&h00C2
PIO program line 1,5,&h4010
PIO program line 1,6,&h8000
PIO program line 1,7,&h0045

'configuration
f=1e4 'PIO run at 10kHz
p=Pio(pinctrl 0,0,0,gp0,,,) 'IN base = GP0
e=Pio(execctrl gp22,0,7) 'wrap 7 to 0, gp22 is conditional JMP pin
s=Pio(shiftctrl 0,0,0,0,0,0) 'shift in through LSB, out is not used

'write the configuration, running 10kHz
PIO init machine 1,0,f,p,e,s,0 'start address = 0

'---------------------------- LA code MMBasic ----------------------------------
'prepare trigger signal set to idle (above PIO code needs falling edge trigger)
SetPin gp22,dout
Pin(gp22)=1

'do not start PIO, the DMA will do that.

'reserve data buffers
'The logic analyzer captures 2*length% (32 bit max) samples.
'These are stored in array "data%()" using 2*length% integers (64 bit variables)
'The DMA buffer has 64bit width, is packed with 32bit FIFO (2 per cell) = length%
'The size of the ring buffer is specified in bytes, so size is 8*length%

length%=256

Dim data%(2*length%-1) 'array to put the 32 bit samples (in option base 0)
'Dim packed%(length%-1) 'DMA array to pack 32 bit samples in 64 bit integers
Dim packed% 'name of the array to be formed as circular array
PIO MAKE RING BUFFER packed%,8*length% 'this creates array called "packed%()"

'dim display characters
Dim a$(1)=("_","-")

'---------------------------------------- MAIN -----------------------------------------
Do
Print "press <SPACE> to arm LA"
Do : Loop Until Inkey$=" "

'The DMA should take 2*length samples (= loopback value)
'It should never trigger the ReadyInt interrupt (ringbuffer) hence we make the number of
'samples near infinite (&hFFFFFFFF). When we force stop the DMA then ReadyInt will trigger.
'the number of samples.
PIO DMA_IN 1,0,&h7FFFFFFF,packed%(),ReadyInt,32,2*length% 'start DMA with ring buffer

'write to PIO the number of post trigger samples, this starts the logging of pre-trigger data
PIO WRITE 1,0,1,length%

Pause 200

Print "press 't' to trigger LA"
Do : Loop Until Inkey$="t"

'the trigger pulse, must be wider that 3 / f(pio) in seconds.
'for 10kHz, that is 0.3ms
Pulse gp22,1 'pulse 1ms (1->0->1)

'from here the PIO should autocomplete and wait for another post trigger length% written
Print "sampling post-trigger data"

'wait a bit (can calculate from post trigger length and sampling speed), default for now
Pause 200

'stop DMA, this will call ReadyInt that stops the PIO and re-init it.
PIO DMA_IN OFF

'process the data from the ring buffer
Memory unpack packed%(),data%(),2*length%,32

'calculate position of the trigger point in the ring buffer
endpoint%=Pio(DMA RX POINTER)-Peek(VARADDR packed%()) 'value in bytes
endpoint%=endpoint% / 4 'convert to 32 bit values for use in data%()
trigger%=(endpoint%-length%) 'recalculate trigger from end capture
If trigger%<0 Then trigger%=trigger%+2*length% 'look at wrap around
'Print trigger%

'calculate view window based on screen size (width = 128 chars)
start%=trigger%-64
stop%=trigger%+63

'output to serial port screen
For j=0 To 15 'check 15 bits GP0..GP15
mask%=2^j 'mask GPx
If j=15 Then 'assume GP15 also wired trigger
Print " T ";
Else
Print Right$("0"+Str$(j),2)+" "; 'channel number
EndIf

'now we select 3 cases depending on start and stop. This must be optimized later
'case 1: view window inside array boundaries
'case 2: start% outside array boundaries
'case 3: staop% outside array boundaries

'case 2
If start%<0 Then
For i=start%+2*length% To 2*length%-1
Print a$(((data%(i) And mask%)=mask%));
Next i
For i=0 To stop%
Print a$(((data%(i) And mask%)=mask%));
Next i

'case 3
ElseIf stop%>2*length%-1 Then
For i=start% To 2*length%-1
Print a$(((data%(i) And mask%)=mask%));
Next i
For i=0 To stop%-2*length%
Print a$(((data%(i) And mask%)=mask%));
Next i

'case 1
Else
For i=start% To stop%
Print a$(((data%(i) And mask%)=mask%));
Next i
EndIf
Print
Next j

'clear array (maybe not needed)
Math set 0,packed%()

Loop

End

'-----------------------------------SUBS MMBasic --------------------------------
'stops and re-initializes the PIO to be ready for next DMA
Sub ReadyInt
PIO STOP 1,0
PIO init machine 1,0,f,p,e,s,0 'start address = 0
Print "done"
End Sub

P.S. Peter. Is it better to standardize on RX and TX and not mix with IN and OUT ? Or do you have a reason to use these names ? Standardize on IN and OUT is another option.

PicomiteVGA PETSCII ROBOTS

matherp
Guru

Joined: 11/12/2012
Location: United Kingdom
Posts: 9139

Posted: 11:57am 20 Jan 2023

Copy link to clipboard

Print this post

You mean change DMA_IN and DMA_OUT to DMA_RX and DMA_TX? If so, yes on the next beta

Volhout
Guru

Joined: 05/03/2018
Location: Netherlands
Posts: 4256

Posted: 12:56pm 20 Jan 2023

Copy link to clipboard

Print this post

matherp said You mean change DMA_IN and DMA_OUT to DMA_RX and DMA_TX? �If so, yes on the next beta

Sorry I was not clear: yes that was what I suggested. Either use RX and TX everywhere, or use IN and OUT everywhere. That will make the user manual more clear, and make it more intuitive.

Thanks for your continued support !!

PicomiteVGA PETSCII ROBOTS

matherp
Guru

Joined: 11/12/2012
Location: United Kingdom
Posts: 9139

Posted: 01:03pm 20 Jan 2023

Copy link to clipboard

Print this post

NB: to all

B6 has introduced an error in DMA_OUT loopback which is now ignored. I've found the problem code which is a call to the SDK which should have no affect but does. I'll post b7 with the fix when I've done a bit more investigation
Edited 2023-01-21 00:04 by matherp

matherp
Guru

Joined: 11/12/2012
Location: United Kingdom
Posts: 9139

Posted: 02:08pm 20 Jan 2023

Copy link to clipboard

Print this post

PicoMite(VGA)V5.07.07b7

https://geoffg.net/Downloads/picomite/PicoMite_Beta.zip

Fixed bug in DMA loopback for output.
Fixed formatting bug in GUI output of floats
Command name changes:
DMA RX was DMA_IN
DMA TX was DMA_OUT

Volhout
Guru

Joined: 05/03/2018
Location: Netherlands
Posts: 4256

Posted: 10:00pm 20 Jan 2023

Copy link to clipboard

Print this post

Hi Peter,

I hope you can help me. Attached ZIP has 2 programs. LA_4.bas and LA_5.bas.
The program LA_4.bas is the logic analyzer that works as I expect. If you type "t" repeated, it shows the sampled data from the 15 input ports of a picomite.
Using V50707b7
If you connect GP22 to GP15 you also see the trigger.

Then program LA_5.bas
This program shows corrupted data in the last 128 samples of the data%() array (alternating correct value, next sample 0). Only the last 128 samples.
If you change the value of "samples%" from 512 to 256, or 1024, the last 128 samples show this behaviour (the numerics show the start pointer, mid pointer, and stop pointer of the displayed values).

You may have to type "t" a few times. When start%<0 or stop%>samples%-128 it is visible,

commented out is a section that displays the data that is in the packed%() array. And also the packed%() array shows this artefact. The last 64 values seem corrupted.

And I have no idea what is going wrong. The PIO code is the same, the DMA configuration is the same. The levels of the signals on my breadboard are correct.

I have looked for hours, and cannot find what is going wrong. Have stripped LA_5.bas back to almost identical to LA_4.bas. I have no idea.
I hope you can spot what is going wrong.

Volhout
LA.zip

Edited 2023-01-21 08:09 by Volhout

PicomiteVGA PETSCII ROBOTS

phil99

Guru

Joined: 11/02/2018
Location: Australia
Posts: 2141

Posted: 12:53am 21 Jan 2023

Copy link to clipboard

Print this post

Playing with LA4.bas found an odd thing with PWM Sync. It doesn't start PWM channel 4 and stops it if it was already running.

Here is what was added to LA4.bas

'logic analyzer framework research program
'samples GP0...GP15 with triggering on GP22
'uses PIO and �ring buffer under DMA.
'requires V50707b7 or newer (function names changed)

' Test signal: Three phase motor drive - picomite V50706+
' runs autonomously on GP0...GP5 50Hz

SetPin gp0,pwm �'CH 0a
SetPin gp1,pwm �'CH 0b
SetPin gp2,pwm �'CH 1a
SetPin gp3,pwm �'CH 1b
SetPin gp4,pwm �'CH 2a
SetPin gp5,pwm �'CH 2b
SetPin gp6,pwm �'CH 3a
SetPin gp7,pwm �'CH 3b
SetPin gp8,pwm �'CH 4a
SetPin gp9,pwm �'CH 4b
SetPin gp10,pwm �'CH 5a
SetPin gp11,pwm �'CH 5b

Fpwm = 50
PW = 100 / 6

PWM 0, Fpwm, PW, PW - 100, 1, 1
PWM 1, Fpwm, PW, PW - 100, 1, 1
PWM 2, Fpwm, PW, PW - 100, 1, 1
PWM 3, Fpwm, PW, PW - 100, 1, 1
PWM 4, Fpwm, PW, PW - 100, 1, 1
PWM 5, Fpwm, PW, PW - 100, 1, 1

PWM sync 0, 100/6, 200/6, 300/6, 400/6, 500/6
Pause 20

'----------------------------------- LA code PIO --------------------------

This is the output

Having already been caught by a faulty GP0 (needs a pulldown resistor). I have tested GP8 & 9 to verify that they function properly with DOUT, DIN and PWM.

This also occurs at the command line, not just in the program, which shows exactly what I see on the scope.

Edit
The pin numbers for PWM7 seem to be cross linked to other pins.

PicoMite MMBasic Version 5.07.07b7
Copyright 2011-2023 Geoff Graham
Copyright 2016-2023 Peter Mather

> pwm 6, off
> pwm 7, off
> setpin gp15,pwm
> setpin gp14,pwm
Error : Already Set to pin 20
> setpin gp13,pwm
Error : Already Set to pin 20
> setpin gp12,pwm
> setpin 20, off
> setpin 19, off
> setpin 17, off
> setpin gp14,pwm
> setpin gp15,pwm
Error : Already Set to pin 19
>

Found when using this:-

Fpwm = 50
PW = 100 / 8

For n%=0 To 15 'Set the PWM pins GP0 to GP15
�PINn = MM.Info(pinno "GP"+Str$(n%)):Print "GP";n%,"Pin";PINn
�SetPin PINn, PWM
�If n% And 1 Then PWM n%\2, Fpwm, PW, PW - 100, 1, 1
Next

PWM sync 0, 100/8, 200/8, 300/8, 400/8, 500/8, 600/8, 700/8
Pause 20

'----------------------------------- LA code PIO --------------------------

It works up to PWM6 but not PWM7 - GP14 & GP15.
.
Edited 2023-01-21 14:14 by phil99

Volhout
Guru

Joined: 05/03/2018
Location: Netherlands
Posts: 4256

Posted: 06:54am 21 Jan 2023

Copy link to clipboard

Print this post

Peter,

Thonking about the artefact in LA_5
Looks like from a certain location in the circular buffer the dma is writing fifo data(32-to-64), in stead of packing it(32+32-to-64).
Maybe the location where the buffer is in Ram is the cause.

Or the packed array and data array overlap, the memory unpack overwrites part of the packed array.

Volhout
Edited 2023-01-21 17:52 by Volhout

PicomiteVGA PETSCII ROBOTS

Volhout
Guru

Joined: 05/03/2018
Location: Netherlands
Posts: 4256

Posted: 07:55am 21 Jan 2023

Copy link to clipboard

Print this post

@phill,

Do you have an option set for system i2c. That may reserve gp pins.

Volhout

PicomiteVGA PETSCII ROBOTS

Pluto
Guru

Joined: 09/06/2017
Location: Finland
Posts: 359

Posted: 08:00am 21 Jan 2023

Copy link to clipboard

Print this post

Volhout,
the location where the problems start is &h81 .

matherp
Guru

Joined: 11/12/2012
Location: United Kingdom
Posts: 9139

Posted: 08:42am 21 Jan 2023

Copy link to clipboard

Print this post

Phil

Thanks for the report, cut and paste typo

"enabled=16" should have been "enabled|=16"

Will be fixed in next beta. Now to look at Volhout's issue - might be harder

Page 4 of 16

Print this page